Research On Passenger Flow Forecast Of Metro Network Based On Hybrid Model

The passenger flow prediction of urban rail transit can provide the corresponding train dispatching strategy for the metro operation department and improve the efficiency of metro operation.With the increasing number of metro lines opened and under construction in China,the effective prediction of the changes of passenger flow at the metro stations will directly affect the efficient and orderly operation of the whole network metro system.More importantly,if a subway station fails or an emergency occurs around a subway station,it may cause the congestion of passenger flow and affect the public transport order.Therefore,it is necessary to predict the passenger flow of the whole network metro station,so as to take corresponding measures in time to avoid the impact of emergency events on the public transportation of Metro.But at present,most of the research is aimed at a certain station or some hot spots as the research object,and the research angle of the whole network subway station is used to predict the change of passenger flow is relatively less.In view of the above problems,this paper carries out the research on the prediction of the passenger flow of the whole metro network.The main contents are as follows:First of all,according to the characteristics of randomness,trend,imbalance and regularity of passenger flow of all metro stations in the whole city,the regularity characteristics affecting the change of passenger flow of the whole network and the subway stations are deeply excavated from the perspective of the time of the total network passenger flow,the spatial angle of the total network passenger flow and the distribution of the whole network metro stations.The passenger flow prediction system of the metro station in and out of the network is constructed.The climate data which affect the travel law of citizens are extracted,and the climate data is processed quantitatively,and the climate features are added into the feature system.Secondly,in view of the large amount of interference noise data in the passenger flow of some Metro stations,this paper constructs Kalman filter smooth noise reduction algorithm to smooth the passenger flow,and then combines Kalman filter smooth noise reduction algorithm with passenger flow prediction model to improve the accuracy of passenger flow prediction,and also enhance the anti-interference of hybrid model Stability and robustness.Finally,a new model of passenger flow prediction for metro network is proposed based on hybrid model.Combined with the advantages and disadvantages of different models,different training data and verification data,different model verification methods,different input characteristics and different model parameters are used to effectively integrate Kalman xboost model,lightgbm model,Kalman lightgbm model,catboost model and Kalman smooth lightgbm model to construct mixed passenger flow prediction model.The first mock exam algorithm is used to solve the optimal weight of each single model.The prediction results of mixed passenger flow forecasting model are compared with the prediction results of other passenger flow forecasting models,and the results of comparison show that the mixed passenger flow forecasting model has the best passenger flow prediction performance.